Tool simplifies configuration in machine learning IMUs
“The FSM logic lets users run gesture and motion-recognition algorithms directly in the sensor for always-on convenience and low power consumption,” explained the firm. “The ML core implements classifiers for real-time motion and vibration pattern recognition.”
Matching IMU modules include LSM6DSO and LSM6DSOX, both six-axis.
- LSM6DSO contains logic for up to 16 independent state machines, each generating a hardware interrupt activated by a specific motion pattern
- LSM6DSOX also contains logic to implement multiple-stage decision trees for machine-learning classification.
Power savings come from the IMU performing gesture and motion recognition while its host processor sleeps.
The tool on offer is ‘Unico GUI‘, which configures the IMU’s internal registers as users create their own motion-recognition algorithms.
“The GUI also makes light work of combining the module’s accelerometer and gyroscope data with inputs from additional external sensors, such as a magnetometer,” said ST.
There are three versions:
- STSW-MKI109L for Linux
- STSW-MKI109M for Mac OSX
- STSW-MKI109W for Windows